XML Content Management

Content Management
This chapter has already emphasized that the biggest real benefits for XML publishing come with large, highly structured technical publications, such as regulatory documents, dictionaries, legislation, or maintenance manuals for complex equipment, such as aircraft or weapons systems. Such documents typically have many authors and a formal and often complex editorial review process. These living documents are under continual revision for their entire lives. With a dictionary, for example, lexicographers write individual entries, which will then pass through several stages of editing and approval before being added to the dictionary proper. As corrections and new slipsusage examplescome in, lexicographers will revise an entry and start the entire review process again. At some major projects, the lifespan of the documents is longer than that of the authors: For the Oxford English Dictionary, for example, this process has continued nonstop for almost a century and a half. There's never any concept of a finished book; each edition is only a snapshot of a never-ending work in process. Exactly the same process applies to the aircraft maintenance manual for a large airliner, for example, except that the snapshots are published every few months rather than every few decades.

Obviously, this kind of publishing was possible before computers, much less XML, but it was labor- and paper-file-intensive: precisely the kinds of operations that can benefit from some kind of content-management system. Many computer programmers are already familiar with source-code-management systems, such as CVS or Visual SourceSafe. Content-management systems for documentation are very similar: They allow authors to check objects, such as documents or pictures, into or out of a central repository, which tracks revisions and often also allows searching, indexing, and even final document assembly. A documentation content-management system may also have a workflow component attached to it, so that it can both track the status of each object through the editorial process and manage the process by sending files to the people who need to approve them.

None of these features is unique to XML, but they are especially likely to be required in a large, multiauthor XML documentation project. Some systems go further: Instead of managing each XML document as a single object, the same as a picture or word-processing file, they allow users to check out part of an XML documentsay, the third task in the second chapterwhile other users work on other parts of the same XML document. Typically, the system parses the XML document and converts it into a series of entries in a specialized database, then reconstructs it as XML when needed.

For a large project with multiple authors, a content-management system is often a requirement: The only question is whether to use a traditional system or a special XML-aware one. It is best to approach this problem backward and start with the non-XML solution. Assume that a company is creating a large amount of technical documentation in XML, using a team of authors and editors. At any time, authors will have lists of items assigned to them for writing or revision, and editors will have lists of items assigned to them for editing and approval. The technical documentation itself consists mostly of independent tasks. Listing 3-9 shows a fragment of a simplified DTD.

Listing 3-9. Top Level of a DTD






Most of the authoring items moving through the workflow chain are tasks with their associated graphics; the editors will assign tasks to individual authors, who will then pass the tasks back up to the editors for approval. When a task requires revision, the cycle repeats itself.

Would an XML-aware CMS bring much benefit to a project like this? No. Authors do not need to check out a single step or title or an entire chapter, only tasks. The CMS can handle each task as a separate XML document without needing to know anything about the task's internal markup structure.

To require an XML-aware content-management system, there would have to be no standard unit for authoring, editing, or workflow; the author would need to be able to check out and lock anything from a single list item to the entire document. The author of this book has not yet seen such a requirement in the real world. Every big multiauthor project has standard units of work, whether they are tasks, dictionary entries, or newspaper articles.

Content-management systems also typically offer searching, indexing, and packaging. A full-text search is sometimes sufficient, but because XML markup can provide much more detailed content, a good case can sometimes be made for making the search XML-aware. For more on this point, see Chapter 6.